presented by: nova novita

Cross-Jurisdictional Income Shifting by U.S. Multinationals: Evidence from International Bond Offerings

Presented by:

Nova Novita

Sistematika

Introduction

Placement Incentives

Tax Incentives

We predict that U.S. multinationals are more likely to place international bond offerings through foreign subsidiaries if they have domestic TLCF and/or if the foreign subsidiary is located in a high statutory tax rate country.

Tax Incentives

Non-Tax Factors

Data and Sample

• Obtain data on international bond offerings from databases licensed by the Securities Data Company (SDC).

• Obtain information on the ownership structure of U.S. multinationals from directories of corporate affiliations

• Obtain financial data to compute our tax measures and control variables from the Compustat PST files and annual reports.

Sample• Includes international bond offerings by U.S. multinationals that

are not financial institutions or public utilities denominated in the currencies of Australia, Canada, France, Germany, Italy, Japan, or the United Kingdom.

• Includes bond offerings for all the G-7 countries (except the United States) and Australia.

• The bonds are issued during 1987-1997 either through a foreign subsidiary in that country or through the U.S. parent, with placements by a wholly owned U.S. financing subsidiary attributed to the U.S. parent.

• Our final sample includes 220 firm-year observations, with 103 foreign subsidiary bond offerings and 117 U.S. parent bond offerings

Empirical ModelOur study examines an empirical model of the following general

form:Placement Decision = f (firm-specific tax attributes, differencesin corporate statutory tax rate regimes, and non-tax control variables)

Placement Decision, as one if the international bond offering is issued by a foreign subsidiary (yielding a foreign interest deduction), or zero if the offering is issued by the U.S. parent or its U.S. financing subsidiary (yielding a domestic interest deduction)

We measure the explanatory variables at the end of the prioryear to capture the U.S. multinationals' attributes before the offering and to avoid simultaneity bias.

Tax Variables

One of our measures of firm-specific tax attributes is NOL, which equals one if the U.S. multinational reports a domestic TLCF for tax purposes, or zero otherwise.

Because U.S. multinationals with domestic TLCF have incentives to shift interest deductions to foreign jurisdictions, we predict a positive relation between Placement Decision and NOL.

1. NOL

Tax Variables

Our second measure of firm-specific tax attributes (FTC) reflects the impact of foreign tax credit limitations on the marginal tax benefit of a domestic interest deduction.

U.S. multinationals with foreign tax credit limitations that impair their ability to use domestic interest deductions have incentives to source interest deductions against foreign income

2. FTC

we predict a positive relation between Placement Decision and FTC.

Tax Variables

3. High rate

Our measure of differences in corporate statutory tax rate regimes is a dummy variable, High rate, that equals one for a generally high tax rate country or zero for a moderate tax rate country.

This variable provides a general indicator of how corporatestatutory tax rates in the foreign country compared to U.S. corporate statutory rates over the sample period.

We predict a positive relation between Placement Decision and High rate..

NON-TAX CONTROL VARIABLES

1. Market

We include Market as a control variable to separate country-level tax incentives from differences in the market orientation of the country where the foreign subsidiary is located

Market is a dummy variable that equals one for a market-oriented country, or zero for a bank-oriented country.

Because companies located in market-oriented countries are more likely to use bond offerings as a financing source, we predict a positive relation between Placement Decision and Market.


2. Issue Amount

Issue Amount, measured as the bond amount in millions, controls for potential differences in the debt capacity of the U.S. parent versus the foreign subsidiary.

To the extent foreign subsidiaries have lower relative debt capacities, we predict a negative relation between Placement Decision and Issue Amount.

NON-TAX CONTROL VARIABLES3. Size

Size, measured as the natural log of the U.S. multinationals‘ total assets (in millions), provides an overall control for firm size.

No predict between Placement Decision and Size


Bankruptcy controls for the financial condition of the U.S. multinational using Altman's [1968,1993] Z-Score predictor, with lower values indicating a greater risk of bankruptcy.

4. Bankruptcy

To the extent Bankruptcy proxies for the U.S. parent's ability to obtain favorable interest rates relative to the foreign subsidiary, we predict a negative relation between Placement Decision and Bankruptcy

We report descriptive data and univariate tests of differences by placement decision.

The bond offering terms include interest rate percentage(Interest Rate), years to maturity (Maturity), a private placement with a qualified lender indicator (Private), and a call option indicator (Call Option).

Empirical Results=> Descriptive statistics

• For our measures of firm specific tax attributes, U.S. multinationals using foreign subsidiary bond offerings have a higher proportion of domestic tax-loss carryforwards (NOL) and more binding foreign tax credit limitations (FTC).

• For our measures of country characteristics, we find an insignificant relation for Highrate and a positive effect for Market.

• we find that bond offerings issued through a foreign subsidiary are smaller in amount (IssueAmount).

• U.S. multinationals using foreign subsidiary bond offerings have a greater risk of bankruptcy (Bankruptcy)

Descriptive statisticsOur descriptive statistics for the other bond offering terms

indicate that:

• Foreign subsidiary bond offerings have a higher interest rate percentage (Interest Rate), a longer maturity term (Maturity), and a higher proportion of call options (Call Options).

• Higher interest rates are a non-tax cost associated with foreign subsidiary debt.

• This result is only suggestive because we are aggregating

• International bond offerings over different currencies in each placement decision category.

LOGISTIC REGRESSION RESULTS

• We report the results for two models:

• Model 1 captures country characteristics using our High rate and Market variables, provides tests of both firm specific tax attributes (NOL and FTC) and differences in corporate statutory tax rate regimes (Highrate).

• Model 2 provides additional tests of firm-specific tax attributes after controlling for country effects with separate intercepts.

• Use Canada as the comparison in model 2.

• We generally expect negative coefficients on our country intercepts.

• Goodness-of-fit statistics suggest a "good" fit for our empirical models with correct prediction rates for model 1 ranging from 76.1% to 78.6%, and correct prediction rates for model 2 ranging from 76.9% to 90.3%.


• Model 1 suggest that both firm-specific tax attributes (NOL and FTC) and differences in corporate statutory tax rate regimes (Highrate) affect placement decisions.

• Consistent with our predictions regarding firm-specific tax attributes, we find that:

• U.S. multinationals are more likely to issue international bonds through foreign subsidiaries if they have domestic tax-loss carryforwards.

• the likelihood of foreign subsidiary bond offerings relates positively to the impact of binding foreign tax credit limitations on the marginal tax benefit of domestic interest deductions.

1. For country effects, we find that U.S. multinationals are more likely to place international bond offerings through foreign subsidiaries in "high" tax rate countries.


• For the control variables in model 1, we find the expected positive relation between Placement Decision and Market and the expected negative relation between Placement Decision and Issue Amount.

• These results suggest that U.S. multinationals are more likely to issue bonds through a foreign subsidiary if the subsidiary is located in a market-oriented country, and that they are less likely to issue bonds through a foreign subsidiary as the offering amount increases.

• We also find a positive relation between Placement Decision and Size, and no relation between Placement Decisionand Bankruptcy.

LOGISTIC REGRESSION RESULTS• In model 2, we find consistent results for the firm-specific

tax attributes (NOL and FTC) .

• We also find negative and generally significant coefficients on the country intercepts, which is consistent with the prediction that U.S. multinationals are more likely to use foreign subsidiary bond offerings in Canada than in the other countries in our sample.

• We find similar results for the control variables except that the significance levels are lower for Issue Amount and Size.

SENSITIVITY ANALYSES• Sensitivity analyses indicate that our results are generally robust

to tests of our measures and sample specification.

• Our NOL measure is a dummy variable that indicates whether U.S. multinationals are in domestic tax-loss carryforward positions.

• We conduct sensitivity tests using Compustat-based measure (Graham's [I9961 simulated marginal tax rates).

• we find consistent tax effects with either an indicator measure of firms' domestic tax-loss carryforward positions, or with Graham's continuous measure of simulated marginal tax rates.

SENSITIVITY ANALYSES

• As a sensitivity test of our foreign tax credit limitations result, we exclude observations where FTC is coded one (six of the 220 firm-year observations).

• We find similar results with these observations excluded from the sample.

• As a sensitivity test of our country-level tax result, we use a continuous measure of corporate statutory tax rate differences as an alternative to the Highrate measure.

• We find that our tax results for Highrate, NOL, and FTC are unchanged, and that the English-origin systems report significantly more foreign subsidiary bond offerings than either the German- or French-origin systems.

SENSITIVITY ANALYSES

• As a sensitivity test of our financial condition control, we use the domestic bond rating of the U.S. parent as an alternative to our Bankruptcy control.

• Using the domestic bond rating of the parent company as reported by Moody's, we find similar tax results except that NOL becomes significant at the 0.05 level rather than the 0.01 level

• In our final sensitivity tests, we explore the sample specification: • First, we include the 17 additional firm-year observations for dual

placements, We find similar results with these additional observations included in the sample

• Second, because we use pooled cross-sectional data, we estimate the model with only one country observation for each U.S. multinational.

1. With 90 firm-year observations, yields similar but somewhat less significant results (i.e., FTCremains significant at the 0.01 level, while Highrate and NOLbecome significant at the 0.05 level)

SENSITIVITY ANALYSES• Third, because prior research (Blackwell and Kidwell, I988)

suggests that firms using private offerings to qualified lenders differ from firms using public offerings.

• we estimate both models with the 22firm-year observations for private offerings excluded and find similar results.

Conclusions• Our study provides evidence that tax incentives influence where U.S.

multinationals locate their interest deductions worldwide.

• Consistent with income-shifting predictions, we find that U.S. multinationals are more likely to place bonds through a foreign subsidiary when they have binding foreign tax credit limitations that impair their ability to use domestic interest deductions, or when they are in domestic tax-loss carryforward positions.

• We also find some evidence that U.S. multinationals are more likely to issue bonds through a foreign subsidiary if the subsidiary is located in a country with generally high corporate statutory tax rates compared to the United States.

Conclusions• Our results suggest that U.S. multinationals‘ debt location

decisions take into account the effect ofjurisdiction-specific tax-loss carryforwards and binding foreign tax credit limitations on the value of debt tax shields.

• Our results are also consistent with U.S. multinationals locating interest deductions in different tax jurisdictions as a inechanism to achieve tax-motivated income shifting.

presented by: nova novita

Documents